This work presents an analytical investigation of wave propagation in micropolar thermoelastic plates using fractional-order derivatives to represent memory, nonlocal interactions, and anomalous thermal diffusion. The novelty of the study lies in developing a unified analytical framework that incorporates the Lord–Shulman, Sherief, Youssef, and Ezzat thermoelastic theories and derives modified Rayleigh–Lamb dispersion relations for symmetric and skew-symmetric modes under different thermal boundary conditions. The coupled governing equations for mechanical displacement, microrotation, and temperature are formulated using fractional-order operators. Analytical expressions for the field variables and secular equations are derived for stress-free, couple-stress-free, thermally insulated, and isothermal plate surfaces. The effects of fractional parameters and boundary conditions on phase velocity, attenuation, stress, couple stress, and temperature are then examined. Numerical computations demonstrate clear differences among the four thermoelastic theories. Fractional-order effects significantly modify phase velocity, attenuation, stress amplitudes, microrotation, and temperature distribution compared with the classical Lord–Shulman model. The unified formulation therefore enables direct comparison of different fractional thermoelastic models within the same plate-wave framework. The added value of the present study is the unified treatment and comparative assessment of multiple fractional thermoelastic theories through generalized modified Rayleigh–Lamb equations. The results provide useful guidance for controlling wave propagation and energy dissipation in composites, metamaterials, microscale structures, and other advanced microstructured materials.
This work presents a simplified torque control strategy for achieving optimal braking performance under certain and uncertain road conditions. The approach is based on a two-step torque input: an initial high torque to quickly build braking force, followed by a corrective torque adjusted using a feedback signal. The optimal braking torque is determined through numerical optimization of a quarter-car tire-brake model, accounting for nonlinear dynamics and slip-dependent friction. When road conditions are unknown, the method employs precomputed fitting functions to estimate the friction coefficient and adapt the torque accordingly. Simulation results demonstrate that the proposed strategy maintains near-optimal slip, minimizes braking distance, and ensures system stability even under varying friction scenarios. Compared to the classical anti-lock braking system (ABS) and uncontrolled braking methods, the proposed approach significantly improves braking efficiency and robustness. The low braking-distance overshoot across different speeds and road conditions highlights the method’s effectiveness and suitability for real-time application in advanced braking systems.
Sealless centrifugal canned motor pumps (CMPs) have an enclosed rotating assembly, making direct measurement of rotor motion difficult during routine testing. Vibration assessment therefore relies mainly on measurements from the external pump–motor housing. This study investigates the frequency-domain vibration behaviour of industrial CMPs and develops a case-based framework for interpreting characteristic spectral features. Vibration measurements were performed on three industrial CMPs under steady-state operating conditions using clean water as the pumped fluid and bearing lubricant. Tri-axial accelerometers were mounted on the external housing near the upper and lower hydrodynamic journal bearings. The measured signals were analysed using Fast Fourier Transform (FFT)-based techniques and interpreted using machine design information and established rotordynamic principles. The spectra showed distinct mechanical, hydraulic, bearing-related and electromagnetic frequency regions. A dominant component in the running-speed (1X) region indicated once-per-revolution excitation, with residual rotor unbalance as one possible source. A vane passing frequency (VPF) component was consistent with impeller–volute hydraulic excitation. A subsynchronous component near 0.47X indicated possible fluid-film bearing-related behaviour but was not independently confirmed as fluid whirl. The rotor bar passing frequency (RBPF) and associated sidebands were consistent with electromagnetic excitation. Externally measured housing vibration can support frequency-domain interpretation of industrial CMP behaviour under steady-state conditions. The proposed case-based framework links spectral features with possible excitation mechanisms, verification methods and subsequent engineering actions. The identified features should be treated as engineering indications rather than definitive confirmation of a specific root cause.
Hearing loss affects speech perception and quality of life, creating a need for energy-efficient and low-latency hearing-aid systems with low hardware complexity. The objective of this research is to develop a hardware-efficient speech enhancement and recognition architecture that minimizes area, power consumption, and hardware processing latency while maintaining reliable speech-processing performance for hearing-aid applications. Finite impulse response filters are widely used in hearing aids because of their high performance; however, high-order finite impulse response filters require numerous multiplication and addition operations, leading to increased hardware area and power consumption. To address this issue, this research proposes a memory-less, multiplier-less distributed arithmetic-based finite impulse response filter integrated with a deep spectro neural network optimized using an approximate wallace tree multiplier for speech recognition. The major contribution of this research is the unified integration of hardware-efficient speech enhancement and neural-network inference within a single application-specific integrated circuit-oriented architecture. MATLAB is used for functional validation, whereas the hardware architecture is evaluated through application-specific integrated circuit post-synthesis analysis, demonstrating an implementation area of 160,000 µm2, power consumption of 1.4 mW, an area-delay product of 1,321,222 µm2·ns, a power-delay product of 11.56 mW·ns, and a maximum throughput of 121.1 MHz corresponding to a system critical-path delay of approximately 8.26 ns. The hardware processing latency is approximately 2.1 ms after speech-frame acquisition. These findings confirm the effectiveness of the proposed architecture for energy-efficient hearing-aid systems. The current research is limited to MATLAB-based functional validation and application-specific integrated circuit post-synthesis analysis, and future research will focus on prototype fabrication and real-time hardware validation.
Excessive vibrations in wind turbine gearboxes significantly affect reliability and service life, making their mitigation a critical issue in vibration engineering. This study investigates the parametric reduction of gearbox vibration using a nonlinear lumped-parameter dynamic model combined with data-driven surrogate-assisted optimization. A nonlinear lumped-parameter dynamic model of a parallel-shaft spur gear reducer is developed, explicitly accounting for parametric excitation induced by time-varying mesh stiffness. Six mechanically relevant design parameters are varied within physically realistic bounds to explore their influence on the system dynamic response. A database of 729 transient simulations is generated from the physical model and used to construct a surrogate representation of the vibration response. Several regression approaches are compared, and the selected XGBoost surrogate is coupled with a Bayesian-tuned particle swarm optimization procedure. Optimization repeatability, independent genetic-algorithm cross-verification, and design-space sensitivity analysis are used to assess the robustness of the solution. XGBoost provides the highest predictive accuracy among the evaluated regression approaches, with an RMSE of 5.90 × 10⁻³, an MAE of 4.19 × 10⁻³, and an R² of 0.9995. Full-model re-evaluation of the optimized configuration shows vibration amplitude reductions ranging from 55.90
